Ready‐to‐use therapeutic food for home‐based treatment of severe acute malnutrition in children from six months to five years of age
Abstract
Background
Malnourished children have a higher risk of death and illness. Treating severe acute malnourished children in hospitals is not always desirable or practical in rural settings, and home treatment may be better. Home treatment can be food prepared by the carer, such as flour porridge, or commercially manufactured food such as ready‐to‐use therapeutic food (RUTF). RUTF is made according to a standard, energy‐rich composition defined by the World Health Organization (WHO). The benefits of RUTF include a low moisture content, long shelf life without needing refrigeration and that it requires no preparation.
Objectives
To assess the effects of home‐based RUTF on recovery, relapse and mortality in children with severe acute malnutrition.
Search methods
We searched the following electronic databases up to April 2013: Cochrane Central Register of Clinical Trials (CENTRAL), MEDLINE, MEDLINE In‐process, EMBASE, CINAHL, Science Citation Index, African Index Medicus, LILACS, ZETOC and three trials registers. We also contacted researchers and clinicians in the field and handsearched bibliographies of included studies and relevant reviews.
Selection criteria
We included randomised and quasi‐randomised controlled trials where children between six months and five years of age with severe acute malnutrition were treated at home with RUTF compared to a standard diet, or different regimens and formulations of RUTFs compared to each other. We assessed recovery, relapse and mortality as primary outcomes, and anthropometrical changes, time to recovery and adverse outcomes as secondary outcomes.
Data collection and analysis
Two review authors independently assessed trial eligibility using prespecified criteria, and three review authors independently extracted data and assessed trial risk of bias.
Main results
We included four trials (three having a high risk of bias), all conducted in Malawi with the same contact author. One small trial included children infected with human immunodeficiency virus (HIV). We found the risk of bias to be high for the three quasi‐randomised trials while the fourth trial had a low to moderate risk of bias. Because of the sparse data for HIV, we reported below the main results for all children together.
RUTF meeting total daily requirements versus standard diet
When comparing RUTF with standard diet (flour porridge), we found three quasi‐randomised cluster trials (n = 599). RUTF may improve recovery slightly (risk ratio (RR) 1.32; 95% confidence interval (CI) 1.16 to 1.50; low quality evidence), but we do not know whether RUTF improves relapse, mortality or weight gain (very low quality evidence).
RUTF supplement versus RUTF meeting total daily requirements
When comparing RUTF supplement with RUTF that meets total daily nutritional requirements, we found two quasi‐randomised cluster trials (n = 210). For recovery, relapse, mortality and weight gain the quality of evidence was very low; therefore, the effects of RUTF are unknown.
RUTF containing less milk powder versus standard RUTF
When comparing a cheaper RUTF containing less milk powder (10%) versus standard RUTF (25% milk powder), we found one trial that randomised 1874 children. For recovery, there was probably little or no difference between the groups (RR 0.97; 95% CI 0.93 to 1.01; moderate quality evidence). RUTF containing less milk powder may lead to slightly more children relapsing (RR 1.33; 95% CI 1.03 to 1.72; low quality evidence) and to less weight gain (mean difference (MD) ‐0.5 g/kg/day; 95% CI ‐0.75 to ‐0.25; low‐quality evidence) than standard RUTF. We do not know whether the cheaper RUTF improved mortality (very low quality evidence).
Authors' conclusions
Given the limited evidence base currently available, it is not possible to reach definitive conclusions regarding differences in clinical outcomes in children with severe acute malnutrition who were given home‐based ready‐to‐use therapeutic food (RUTF) compared to the standard diet, or who were treated with RUTF in different daily amounts or formulations. Well‐designed, adequately powered pragmatic randomised controlled trials of HIV‐uninfected and HIV‐infected children with severe acute malnutrition are needed.
PICOs
Plain language summary
Ready‐to‐use therapeutic food as home‐based treatment for severely malnourished children between six months and five years old
Malnourished children have a higher risk of death and illness. Treating severely malnourished children in hospitals is not always desirable or practical in rural settings, and home treatment may be better. Home treatment can be food prepared by the carer, such as flour porridge, or commercially manufactured food such as ready‐to‐use therapeutic food (RUTF). RUTF is made according to a standard, energy‐rich composition defined by the World Health Organization. Typically, RUTF is made from full‐fat milk powder, sugar, peanut butter, vegetable oil, and vitamins and minerals. The benefits of RUTF include a low moisture content, a long shelf life without needing refrigeration and that it requires no preparation.
We assessed RUTF compared with a standard diet (flour porridge) for treatment, and examined whether a cheaper RUTF treatment (smaller amounts or using cheaper ingredients) can achieve similar health outcomes in severely malnourished children between six months and five years old. The main health outcomes that we investigated were recovery from severe malnutrition, relapse (getting more malnourished), death and weight gain.
We carried out a comprehensive search of trials up to April 2013 and found four studies. All studies were conducted in Malawi, with one small study that included children infected with human immunodeficiency virus (HIV). The extent to which results of the studies can be believed based on how the studies were done was poor for three studies, while the fourth study had stronger methods. Because of the sparse data for HIV, we report the main results for all children together.
For RUTF given as a total dietary replacement compared to flour porridge, we found three studies with 599 children. RUTF may improve recovery slightly, but we do not know whether RUTF improves relapse, death or weight gain as the quality of evidence was very low.
When comparing RUTF used as a supplement to their ordinary diet with RUTF used as a total dietary replacement, we found two small studies with 210 children. For recovery, relapse, death and weight gain, the quality of evidence was very low and, therefore, we do not know what the effects are.
When comparing a cheaper RUTF containing less milk powder (10%) with standard RUTF (25% milk powder), we found one study that randomised 1874 children. For recovery, there probably was little or no difference between the groups. RUTF containing less milk powder may lead to slightly more children relapsing and to less weight gain than standard RUTF. We do not know whether the cheaper RUTF reduces the number of children dying.
Current evidence is limited and, therefore, we cannot conclude that there is a difference between RUTF and flour porridge as home treatment for severely malnourished children, or between RUTF given in different daily amounts or with different ingredients. In order to determine the effects of RUTF, more high‐quality studies are needed.
Authors' conclusions
Summary of findings
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Standard diet | RUTF | |||||
| Recovery | 597 per 1000 | 788 per 1000 | RR 1.32 | 599 | ⊕⊕⊝⊝ | ‐ |
| Relapse | See comment | See comment | Not estimable | 599 | ⊕⊝⊝⊝ | 2 studies found a large effect with RUTF, 1 study did not detect an effect5 |
| Mortality | 54 per 1000 | 53 per 1000 | RR 0.97 | 599 | ⊕⊝⊝⊝ | ‐ |
| Weight gain | ‐ | The mean weight gain in the intervention groups was | ‐ | 595 | ⊕⊝⊝⊝ | ‐ |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The three studies had different definitions for recovery. Ciliberto 2005: reaching WHZ > ‐2; Manary 2004: reaching WHZ ≥ 0; Ndekha 2005: reaching 100% weight for height. | ||||||
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| RUTF (total daily requirements) | RUTF supplement | |||||
| Recovery | 830 per 1000 | 589 per 1000 | RR 0.71 | 210 | ⊕⊝⊝⊝ | ‐ |
| Relapse | 0 per 1000 | 0 per 1000 | RR 8.95 | 210 | ⊕⊝⊝⊝ | ‐ |
| Mortality | 68 per 1000 | 50 per 1000 | RR 0.73 | 210 | ⊕⊝⊝⊝ | ‐ |
| Weight gain: HIV‐uninfected children | ‐ | The mean weight gain: HIV‐uninfected children in the intervention groups was | ‐ | 158 | ⊕⊝⊝⊝ | Weight gain in g/kg/day for a 6‐kg child translates to a clinically important difference of MD 378 g/month (95% CI 201.6 to 554.4) |
| Weight gain: HIV‐infected children | ‐ | The mean weight gain: HIV‐infected children in the intervention groups was | ‐ | 48 | ⊕⊝⊝⊝ | Weight in g/kg/day for a 6‐kg child translates to a clinically important difference of either a loss of 311.4 g/month or a gain of 275.4 g/month |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The 2 studies used different definitions for recovery. Manary 2004: reaching WHZ ≥ 0; Ndekha 100% weight for height. | ||||||
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Standard RUTF | RUTF containing less milk powder | |||||
| Recovery | 836 per 1000 | 811 per 1000 | RR 0.97 | 1874 | ⊕⊕⊕⊝ | ‐ |
| Relapse | 98 per 1000 | 131 per 1000 | RR 1.33 | 1874 | ⊕⊕⊝⊝ | ‐ |
| Mortality | 36 per 1000 | 32 per 1000 | RR 0.90 | 1874 | ⊕⊝⊝⊝ | ‐ |
| Weight gain | ‐ | The mean weight gain in the intervention groups was | ‐ | 1874 | ⊕⊕⊝⊝ | Weight in g/kg/day for a 6‐kg child translates to a clinically unimportant difference of a loss of MD 90 (95% CI 135 to 45) g/month |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The intervention period was up to 8 weeks, but outcome data for children who recovered, relapsed or died before 8 weeks was captured until the time the child was finished with the study. | ||||||
Background
Description of the condition
Malnutrition occurs when the quantity of one or more macronutrients available to body tissues is inadequate to sustain optimal bodily functions (Manary 2008), and this is usually accompanied by numerous micronutrient deficiencies. Malnutrition is a broad concept that includes a variety of clinical conditions such as kwashiorkor, marasmus, marasmic kwashiorkor, wasting or stunting, and micronutrient deficiencies. For the purpose of this review, the term malnutrition only refers to undernutrition. Macronutrient malnutrition is the focus of this review, so it includes all of the above conditions, which could also be accompanied by different degrees of micronutrient deficiencies.
Malnutrition commonly affects infants and young children, pregnant and lactating women, and elderly people. More than 77 million children are born every year in the 36 countries with the highest burden of malnutrition (21 of these countries are in Africa, 13 in Asia and two in Latin America) (Bhutta 2008; Black 2008). Of these children, about 7.4 million die before the age of three years and a further 0.6 million die between the ages of three and five years (Bhutta 2008). Short‐term consequences of malnutrition include mortality and morbidity, for example, pneumonia, diarrhoea, fatigue and impaired thermoregulation (Black 2008). In the long term, malnutrition in children may affect adult size, intellectual ability, economic productivity and reproductive performance, and increase the risk of metabolic disorders and cardiovascular disease (Black 2008).
In children under five years of age, malnutrition can be classified as moderate or severe. Moderate malnutrition ‐ often referred to as moderate acute malnutrition (MAM) ‐ is defined as a weight for height z score (WHZ) between two and three standard deviations (SDs) below the mean. Severe malnutrition ‐ often referred to as severe acute malnutrition (SAM) ‐ is defined as a WHZ of more than three SDs below the mean, or a mid‐upper arm circumference (MUAC) of less than 115 mm, or the presence of nutritional oedema (Collins 2003; Manary 2008; WHO and UNICEF 2009). MAM or SAM without bilateral pitting oedema is termed marasmus. In the presence of bilateral pitting oedema, the term kwashiorkor is used (Manary 2008). See Table 1 for a more detailed classification system for MAM and SAM.
| Severe acute malnutrition with complications | Severe acute malnutrition without complications |
| Bilateral pitting oedema grade 3*** (severe oedema) OR | MUAC < 110† mm OR |
| MUAC < 110† mm and bilateral pitting oedema grades 1* or 2** (marasmic kwashiorkor) OR | Bilateral pitting oedema grades 1* or 2** with MUAC ≥ 110† mm AND
|
| MUAC < 110† mm or bilateral pitting oedema grades 1* or 2** AND 1 of the following:
| ‐ |
| ↓ | ↓ |
| Inpatient care IMCI/WHO protocol | Outpatient therapeutic care protocols |
IMCI: Integrated Management of Childhood Illness; MUAC: mid‐upper arm circumference; UNICEF: United Nations Children's Fund; WHO: World Health Organization.
*Grade 1 = mild oedema on both feet or both ankles.
**Grade 2 = moderate oedema on both feet, and on lower legs, hands or lower arms.
***Grade 3 = severe generalised oedema affecting feet, legs, hands, arms and face.
†The WHO and UNICEF now recommend that the cut‐off value for the MUAC for severe acute malnutrition should be increased to 115 mm (WHO and UNICEF 2009). The adoption of this higher cut‐off value will sharply increase the caseloads, which may influence the cost of nutrition programmes greatly (WHO and UNICEF 2009). However, detecting more children earlier as severely malnourished will lead to a shorter period needed to treat them, which may bring down the cost per child (WHO and UNICEF 2009).
¤IMCI criteria: 60 respirations/min children age < 2 months; 50 respirations/min for age 2‐12 months; 40 respirations/min for age 1‐5 years; 30 respirations/min for age > 5 years.
Although some conditions may contribute to the onset of malnutrition (for example, human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS), tuberculosis (TB), kidney failure), poverty and food insecurity are major causes. Malnutrition and infection have a reciprocal effect since a lower host response to infection contributes to compromised nutritional status and vice versa (Kruger 2008; Naude 2008). Infections are associated with anorexia (loss of appetite) and decreased food intake; fever increases energy expenditure; and diarrhoea decreases nutrient absorption; these result in wasting and higher mortality from infectious diseases (Kruger 2008; Naude 2008).
Description of the intervention
Ultimately, the only way to end malnutrition is to address economic deprivation and inequity. However, conditions can be mitigated by offering specific nutritional interventions (Black 2008). Hospitalised treatment for SAM children typically entails treatment with F75 (the starter milk‐based therapeutic formula; thus referred to as phase 1 or stabilisation phase) (ACF International Network 2009; WHO and UNICEF 2009). During this stabilisation phase, the oedema (if present) starts to disappear, leading to weight loss (fluid loss). F75 aids in boosting the metabolism and restoring hydroelectric equilibrium (ACF International Network 2009). Next, F100 (a milk‐based therapeutic diet; also called phase 2 treatment) is given to initiate weight gain.
Ready‐to‐use foods (RUF) are energy‐dense food with a low moisture content that can be eaten directly from the packaging. When used for nutritional rehabilitation of children with SAM, such products are referred to as ready‐to‐use therapeutic food (RUTF). RUTF was originally developed as a home‐based alternative to F100. RUTF, in the form of a solid or semi‐solid feed, has a similar nutrient profile to F100 (except for the presence of iron) (Collins 2006; WHO 2007). Table 2 shows the nutritional contents of RUTF as recommended by the World Health Organization (WHO).
| Moisture content | 2.5% maximum |
| Energy | 520‐550 kCal/100 g |
| Protein | 101‐2% total energy |
| Lipids | 45‐60% total energy |
| Sodium | 290 mg/100 g maximum |
| Potassium | 1110‐1400 mg/100 g |
| Calcium | 300‐600 mg/100 g |
| Phosphorus (excluding phytate) | 300‐600 mg/100 g |
| Magnesium | 80‐140 mg/100 g |
| Iron | 10‐14 mg/100 g |
| Zinc | 11‐14 mg/100 g |
| Copper | 1.4‐1.8 mg/100 g |
| Selenium | 20‐40 μg |
| Iodine | 70‐140 μg/100 g |
| Vitamin A | 0.8‐1.1 mg/100 g |
| Vitamin D | 15‐20 μg/100 g |
| Vitamin E | 20 mg/100 g minimum |
| Vitamin K | 15 to 30 μg/100 g |
| Vitamin B1 | 0.5 mg/100 g minimum |
| Vitamin B2 | 1.6 mg/100 g minimum |
| Vitamin C | 50 mg/100 g minimum |
| Vitamin B6 | 0.6 mg/100 g minimum |
| Vitamin B12 | 1.6 μg/100 g minimum |
| Folic acid | 200 μg/100 g minimum |
| Niacin | 5 mg/100 g minimum |
| Pantothenic acid | 3 mg/100 g minimum |
| Biotin | 60 μg/100 g minimum |
| n‐6 fatty acids | 3‐10% of total energy |
| n‐3 fatty acids | 0.3‐2.5% of total energy |
RUTF can either be commercially produced on large scale or produced locally (usually on small scale with ingredients that may differ slightly from commercially produced RUTF as these may be locally sourced). Two examples of commercially produced RUTF are a peanut‐based paste called Plumpy'nut® (developed by Nutriset, Plumpy'nut, and the Institute for Research and Development, France) and a solid biscuit made from cooked wheat called BP100® (developed by Compact, Denmark) (Collins 2004; Navarro‐Colorado 2005). Both are fortified with micronutrients and have very low water activity, which discourages microbial growth (Brewster 2006; WHO 2007; Kruger 2008). This is an important feature since clean safe water is not widely available in many poor communities. Children as young as six months can consume RUTF with a homogenous paste texture. Solid RUTF can be soaked in clean boiling water and eaten as porridge by such young children, or older children can consume it as a biscuit.
Communities can also learn how to produce their own RUTF, as in Malawi where a peanut‐based RUTF is produced (Sandige 2010 [pers comm]). Table 3 shows a typical recipe for a peanut‐based RUTF. Examples of other countries that manufacture RUTF are Ethiopia, Niger and the Democratic Republic of Congo in Africa, as well as Sri Lanka, Indonesia and Pakistan in Asia (DFID 2009). The manufacturing equipment and technology needed to produce RUTF is simple and can be transferred to any country with minimal industrial infrastructure (WHO 2007). The methods of quality control that are needed and the exact cost are determined by the scale of production (Manary 2006), but on average RUTF costs approximately USD3 per kilogram when locally (non‐commercially) produced (WHO 2007). In April 2012, we communicated with Nutriset and obtained their prices for Plumpy'nut®: EUR2.7 per kilogram (EUR0.25 for a 92g packet), which excludes transport and import tax cost. Children with SAM normally need 10 to 15 kg of RUTF given over a period of six to eight weeks for recovery from undernutrition (WHO 2007). Authors of a cost analysis in Ethiopia reported that the cost of commercially produced RUTF per child treated at home is USD128 while in a healthcare facility it is significantly more expensive at USD262 (Tekeste 2012). In an analysis conducted in Zambia, the authors reported a cost of USD202 per child for home‐based RUTF treatment when compared to no treatment (Bachmann 2009). Studies comparing the costs of RUTF with standard home‐based treatment (for example, a porridge made from a maize and soy flour blend) are not available.
| Ingredient | % weight |
| Full‐fat milk | 30 |
| Sugar | 28 |
| Vegetable oil | 15 |
| Peanut butter* | 25 |
| Mineral‐vitamin mix | 1.6 |
*Strict quality control is essential.
Recipes for RUTF do not necessarily include peanut or milk powder, although the WHO recommends that at least half of the proteins should come from a milk source (WHO 2007). Peanuts can cause allergic reactions in susceptible individuals and are known to have a high risk for aflatoxin contamination. Milk powder is expensive and often has to be imported (Collins 2004). The cost of milk powder in Malawi constitutes more than half of the cost of the final RUTF (Collins 2004). For non‐commercial production of RUTF, the following basic ingredients should be present (Collins 2004).
-
A staple food as the main ingredient (preferably a cereal).
-
A protein supplement from a plant or animal food (for example, beans, groundnuts, milk, meat, chicken, fish, egg). To make the production of RUTF cost‐effective, legumes and oilseeds are mostly used.
-
A vitamin and mineral supplement (a vegetable or fruit, or both).
-
An energy supplement (a fat, oil or sugar) to increase the energy density.
The food safety of the production process should be strictly monitored, with careful attention given to avoid contamination by microorganisms or other harmful substances (for example, heavy metals, pesticides, anti‐nutritional factors such as phytate or protease inhibitors) (WHO 2007). Table 4, Table 5 and Table 6 give three recipes for locally produced RUTF. Table 7, Table 8 and Table 9 provide nutritional information and water activity of these recipes as well as for Plumpy'nut®.
| Ingredient | Quantities (%) |
| Roasted rice flour | 20 |
| Soyamin 90 | 29 |
| Roasted sesame seed paste | 8 |
| Sunflower oil | 19.4 |
| Icing sugar | 22 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
RUTF: ready‐to‐use therapeutic food.
| Ingredient | Quantities (%) |
| Roasted pearl barley flour | 15 |
| Soyamin 90 | 9 |
| Roasted sesame seed paste | 27 |
| Sunflower oil | 24 |
| Icing sugar | 23.4 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
RUTF: ready‐to‐use therapeutic food.
| Ingredient | Quantities (%) |
| Roasted maize flour | 33.4 |
| Roasted sesame seed paste | 27 |
| Roasted chick pea flour | 25 |
| Sunflower oil | 12 |
| Icing sugar | 15 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
RUTF: ready‐to‐use therapeutic food.
| Nutrient | Unit | RUTF‐1 (100 g) | Energy (%) | RUTF‐2 (100 g) | Energy (%) | RUTF‐3 (100 g) | Energy (%) | Plumpy'nut®* (100 g) | Energy (%) |
| Energy** | kCal | 551 | ‐ | 567 | ‐ | 512 | ‐ | 530 | ‐ |
| Energy | kJ | 2307 | ‐ | 2373 | ‐ | 2142 | ‐ | 2218 | ‐ |
| Protein | g | 13.8 | 10 | 14.1 | 10 | 13.4 | 11 | 14.5 | 11 |
| Carbohydrate*** | g | 43 | 31 | 39.9 | 28 | 50.2 | 39 | 43 | 32 |
| Fat | g | 36 | 59 | 39 | 62 | 28.6 | 50 | 33.5 | 57 |
| Ash | g | 43 | ‐ | 3.9 | ‐ | 4.9 | ‐ | 4 | ‐ |
| Moisture | g | 2.9 | ‐ | 3.1 | ‐ | 2.9 | ‐ | < 5 | ‐ |
*Protein and fat are reported to contribute 11% and 57% in energy input, respectively. Total energy is reported to be 530 kcal/100 g and moisture < 5%.
**The energy has been calculated using Atwater factors.
***Carbohydrate is by difference assuming protein to be nitrogen multiplied by 6.25.
| Mineral | RUTF‐1 (mg/kg) | RUTF‐2 (mg/kg) | RUTF‐3 (mg/kg) | Plumpy'nut®(mg/kg) |
| Cu | 2.1 | 2.1 | 1.8 | 1.7 |
| Zn | 10.9 | 10.9 | 10.2 | 13 |
| Ca | 338.1 | 338.1 | 209.8 | 310 |
| Na | 256.5 | 256.5 | 189.9 | < 290 |
| Mg | 118.4 | 118.4 | 119.1 | 86 |
| Fe | 5.6 | 5.6 | 4.4 | 12.45 |
| RUTF recipe | Water activity |
| RUTF‐1 | 0.290 |
| RUTF‐2 | 0.279 |
| RUTF‐3 | 0.260 |
| Plumpy'nut® | 0.241 |
How the intervention might work
Adequate energy, protein and micronutrient intake is vital for maintaining a functioning immune system or restoring a system that is dysfunctional (Naude 2008). Malnutrition is a condition where the body is in great need of nutrients. Individuals recovering from malnutrition require relatively large amounts of nutrients, in particular energy. Infants and young children have a small body size, which limits the amount of food that can be given in a single feed (Lin 2008). Lower energy‐density foods, together with a low frequency of feeding, can result in an energy intake that is insufficient to enable recovery.
The following characteristics of RUTF may contribute to its possible beneficial effect in the treatment of malnutrition.
-
Balanced, nutritious, home‐based therapy.
-
Affordable compared to facility‐based care.
-
Can be eaten safely at home, even where hygienic conditions are poor (WHO 2007).
-
Long shelf life.
-
No special storage (for example, refrigeration) or preparation required.
Why it is important to do this review
The vast majority of children with malnutrition live in low‐ and middle‐income countries (LMIC). Many of these children never visit healthcare facilities (WHO 2007; Black 2008) due to reasons such as a lack of money for transport to facilities or long travel distances, or both; parents' lack of health status awareness; and a lack of healthcare resources to treat thousands of malnourished children in facilities (Kruger 2008). Furthermore, hospital admission exposes people with uncomplicated SAM to additional risks of nosocomial infections and takes the mother or carer away from other children for prolonged periods, which may increase the risk for sibling malnutrition (Collins 2003). Therefore, an alternative treatment for severe uncomplicated malnutrition may be a home‐based nutritional intervention, such as RUTF, which does not require specialised healthcare personnel and expensive equipment (Kruger 2008).
Both the WHO and the United Nations Children's Fund (UNICEF) now recommend the use of RUTF in the community as therapeutic feeding for children with SAM (WHO and UNICEF 2009) (see Table 10). The findings of this systematic review will be of significant value to people in LMIC as well as to organisations involved in preparing clinical guidelines for practitioners and policy makers in LMIC (for example, WHO, UNICEF and government health departments).
| Independent additional criteria |
|
|
| ↓ | ↓ | |
| Type of therapeutic feeding | Facility‐based | Community‐based |
| Intervention |
|
|
| Discharge criteria (transition criteria from facility to community‐based care) |
|
|
RUTF: ready‐to‐use therapeutic food; UNICEF: United Nations Children's Fund; WHO: World Health Organization.
*Children who eat at least 75% of their calculated RUTF ration for the day.
Objectives
To assess the effects of home‐based RUTF on recovery, relapse, mortality, time to recovery and anthropometrical changes in children with SAM. Specific comparisons investigated were:
-
RUTF meeting total daily nutritional requirements versus standard diet (for example, flour porridge);
-
RUTF supplement versus RUTF meeting total daily nutritional requirements;
-
RUTF containing less milk powder versus standard RUTF.
Methods
Criteria for considering studies for this review
Types of studies
Randomised controlled trials (RCTs), including those defined as quasi‐randomised (that is, trials that used an inadequate method of randomisation, such as alternation or date of birth). We also included cluster randomised trials (that is, trials randomised by groups such as schools, villages or families).
Types of participants
Children between six months and five years of age with SAM, regardless of country, setting or disease status and irrespective of the method of diagnosis employed.
Types of interventions
Experimental
-
RUTF as defined by the study authors (either commercially or non‐commercially produced).
Control
-
Alternative RUTF type (for example, corn/soy‐based versus peanut‐based, reduced milk powder content).
-
Treatment as usual (for example, standard diet).
Any trial in which the effects of an RUTF were potentially confounded by another intervention was excluded, that is, where multiple interventions were involved, comparison groups should have received the same treatment apart from the experimental RUTF.
Types of outcome measures
Primary outcomes
-
Recovery as defined by the study authors.
-
Deterioration or relapse during and beyond the intervention period as defined by the study authors.
-
Mortality.
Secondary outcomes
-
Mean weight gain per kilogram body weight per day during the intervention period.
-
Time to recovery (duration to rehabilitation).
-
Anthropometrical status at all reported time points during and beyond the intervention period (for example, WHZ, weight for age z score (WAZ), height for age z score (HAZ), MUAC).
-
Cognitive function and development during the intervention period (for example, Denver Developmental Screening Test, Bayley Scales of Infant Development).
-
Adverse outcomes as reported by investigators (for example, allergic reactions, refusal of feeds due to poor palatability, diarrhoea).
Search methods for identification of studies
We used a comprehensive search strategy to identify all relevant studies regardless of language or publication status (published, unpublished, in press and in progress).
Electronic searches
We searched the following electronic sources up to April 2013 to identify relevant studies that assessed the effects of RUTF on malnutrition.
-
The Cochrane Central Register of Clinical Trials (CENTRAL), 2013, Issue 3, searched 4 April 2013
-
Ovid MEDLINE 1946 to March week 4 2014, searched 4 April 2013
-
EMBASE 1980 to 2013 week 13, searched 4 April 2013
-
African Index Medicus, searched 8 April 2013
-
CINAHL 1937 to current, searched 8 April 2013
-
Science Citation Index 1970 to 3 April 2013, searched 4 April 2013
-
LILACS, searched 8 April 2013
-
ZETOC (limited to conference search), searched 8 April 2013
-
ClinicalTrials.gov (clinicaltrials.gov/), searched 8 April 2013
-
Current Controlled Trials (www.controlled‐trials.com/), searched 8 April 2013
-
WHO International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch/), searched 8 April 2013
Appendix 1 shows the complete search histories for each database.
Searching other resources
In order to obtain additional references, we contacted researchers, paediatricians and community dieticians. We scrutinised the reference lists of included studies and appropriate reviews in order to identify relevant studies. We contacted the authors of each trial identified in the trial registries to establish whether the trial had already been published, and the authors of all included studies to determine if they were aware of additional trials (published, unpublished or ongoing) in the field.
Data collection and analysis
Selection of studies
Two review authors (AS and ML) independently screened the title and abstract of studies identified by the search and applied the prespecified eligibility criteria in order to identify relevant studies. Where at least one review author considered a study to be relevant we obtained the full text and independently assessed it for eligibility. We contacted the authors of the primary studies where there was missing information or clarification was needed. We resolved any remaining disagreements by consensus among the review authors. We listed studies first thought to be relevant but which we later excluded in the Characteristics of excluded studies with reasons for exclusion.
Data extraction and management
Three review authors (AS, ML and AM) independently extracted data using a standardised, pre‐piloted data extraction form and resolved disagreements by consensus among the review authors. For each relevant study, we extracted the following: source (for example, contact details and citation); methods (for example, ethics approval and study design); participants (for example, age and comorbidity); interventions (for example, description, dose and duration); outcomes (for example, description and time point(s) collected); results (for example, number of participants randomised per arm and numerical results for prespecified outcomes); safety (for example, number and description of adverse effects or events per arm) and miscellaneous information (for example, funding source and references to other relevant studies).
We contacted the study authors where reported information was unclear or contradictory, or where important data were missing.
We entered the extracted data into one of the following three tables: (1) Characteristics of included studies, (2) Characteristics of excluded studies and (3) Characteristics of ongoing studies.
Assessment of risk of bias in included studies
Three review authors (AS, ML and AM) independently assessed each included study for risk of bias using the guidelines provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008) using the specific criteria in Appendix 2. The assessed domains were adequate sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting and other potential sources of bias. We rated each included study as low risk of bias, high risk of bias or unclear risk of bias according to each of the six domains. We discussed disagreements with a fourth review author (JV).
We evaluated cluster‐randomised trials according to the following criteria: recruitment bias, baseline imbalance and loss of clusters (Higgins 2008), using the specific criteria shown in Appendix 3.
Measures of treatment effect
We used Review Manager 5 (RevMan) to manage the data and to conduct the analysis (RevMan 2011). We calculated risk ratios (RR) for dichotomous data and mean differences (MD) for continuous data. We presented all results with 95% confidence intervals (CI).
Unit of analysis issues
Because of the nature of SAM, we did not expect to find any cross‐over trials.
For cluster‐randomised trials, we followed the method of adjusting for clustering as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). None of the three included cluster‐randomised trials had properly accounted for the cluster design. Therefore, we used an 'approximate method', which entailed calculation of an 'effective sample size' for the comparison groups by dividing the original sample size by the 'design effect', which is 1 + (c‐1)ICC, where c is the average cluster size and ICC is the intracluster correlation coefficient. For dichotomous data, we divided both the number of participants and the number who experienced the event by the same design effect, while for continuous data, only the sample size was reduced (means and SDs were left unchanged). The required information was available for two of the cluster‐randomised trials and we contacted the study author of the third trial to obtain the number of clusters. We imputed a low ICC of 0.001 for two studies because we did not anticipate large between‐cluster variability. The clusters in these studies were either the number of weeks of discharge or the days of discharge in the month. In this way, children from the same community were assessed in the same facility. We imputed a higher ICC of 0.005 for the third study because seven different facilities represented seven clusters. We, therefore, expected a certain degree of between‐cluster variability in this study. Although the values are relatively arbitrary, we preferred to use them to adjust the sample sizes due to the implausibility an ICC of 0. We had initially intended to use the generic inverse variance method in RevMan, but since we had values for the totals, means and SDs per group from each study for continuous data, it became unnecessary to do so.
Two studies had two treatment arms that were compared to the same control (Manary 2004; Ndekha 2005). The one treatment arm received sufficient RUTF to meet daily nutritional requirements whereas the second treatment arm received RUTF supplementary to their habitual diet. Therefore, these two treatments were such that the arms could not be combined into a single pair‐wise comparison (Higgins 2008). As these two different treatment arms relate to different research questions (see the three subsections in Effects of interventions section) they were analysed separately. In Comparison 1, we thus selected the arm that received RUTF in sufficient quantity to meet daily nutritional requirements and compared the results of this arm with the results of the full control group (only considering the ICC). In Comparison 2, we selected the arm that received RUTF as a supplement and compared the results of this arm with the results of the full control group (only considering the ICC).
We reported data of all collection time points during and after the intervention period (follow‐up) as stipulated in the 'Types of outcomes' section (unless otherwise stated). We could not group time points as planned owing to the data obtained (one month or less of RUTF treatment, less than one to two or more months of RUTF treatment and more than two to six months of RUTF treatment). The primary outcomes were either measured at the time of recovery (which varied between participants and such individual data were not reported unless time to recovery was an outcome in the study), or at the end of a predetermined time period. We did not distinguish in the analyses between such studies; however, we reported what time points were used with each outcome in each study.
Dealing with missing data
We classified important missing data per study as (1) pre‐randomisation, (2) immediately post‐randomisation or (3) drop‐outs during the intervention phase, alongside reasons for the absence where these were reported in the article (Table 11). We attempted to obtain essential missing data by contacting the study authors via email. We imputed values for ICC because we could not obtain them from published data.
| Study ID | Participants recruited (n) | Pre‐randomisation attrition (n) | Immediately post‐randomisation attrition (n) | Drop‐outs during the intervention period (n) |
| Ciliberto 2005 | 1178 (includes moderate and severe malnourished children) | 0 | 41 (reasons not reported) | 72 (reasons not reported) |
| Manary 2004 | 452 | 77 refused 93 HIV positive | 0 | 47 were "dropouts" (no reasons reported) and 37 died (no reasons reported) |
| Ndekha 2005 | 93 | 0 | 0 | 11 died (no reasons reported) and 17 drop‐outs (no reasons reported) |
| Oakley 2010 | 1961 | 87 (reasons not reported) | 0 | 51 were lost to follow‐up (no reasons reported other than "those lost were more likely to be younger and marasmic") |
For dichotomous data (for example, recovery and mortality during the intervention period), we used the intention‐to‐treat (ITT) principle to calculate effect sizes for individual studies or to pool more than one study. We assumed that the participants who were loss to follow‐up or dropped out of the study did not experience the event of interest. However, for the outcome 'relapse' we assumed that those who dropped out did not receive any treatment (RUTF or standard diet) and, therefore, experienced the event. Furthermore, when assessing dichotomous outcomes (for example, recovery) at follow‐up (for example, six months after the children initially recovered) we employed the available‐case principle, that is, only those who recovered during the intervention period and came back for follow‐up were assessed as opposed to all children who recovered. We consider it is not plausible to assume that those who did not come back deteriorated. For continuous data, we calculated MDs for studies based on the available‐case principle.
Assessment of heterogeneity
We assessed heterogeneity by visual inspection of forest plots and statistically by means of the Chi2 test for heterogeneity (significance level P value < 0.1). We quantified heterogeneity using the I2 test (Higgins 2002) where I2 values of 50% or more indicated a substantial level of heterogeneity (Higgins 2003).
Assessment of reporting biases
We had planned to assess the likelihood of reporting bias with funnel plots using at least 10 studies per outcome (Higgins 2008). However, we included only four studies.
Data synthesis
We anticipated a high degree of heterogeneity due to the inclusion of children with a variety of conditions related to malnutrition (for example, marasmus, kwashiorkor, stunting) and various types of RUTF (for example, corn/soy‐based RUTF, peanut‐based RUTF). For this reason, we used a random‐effects model to combine the results where appropriate. Where substantial statistical heterogeneity existed, we did not pool study results in a meta‐analysis but reported effect sizes per study separately. We evaluated the quality of evidence using the GRADE (Grading of Recommendations Assessment, Development and Evaluation) tool (Guyatt 2011).
Subgroup analysis and investigation of heterogeneity
We intended to compare the intervention effects across the following subgroups:
-
different types of RUTF products (for example, corn/soy‐based versus peanut‐based RUTF);
-
age of children: 6 to 12 months, as this is the ideal period to start weaning from a milk‐based diet; 13 months to 5 years, as these children consume a mixed diet (mostly not breast milk although the child may still be taking some);
-
children with or without comorbid disease (for example, HIV/AIDS, TB, malaria).
The available data, however, allowed us to conduct subgroup analyses only for HIV status.
Sensitivity analysis
We had planned to perform sensitivity analyses; however, since we only identified four studies, we deemed sensitivity analyses inappropriate. In future updates it may be feasible to assess the influence of study quality (using adequacy of allocation concealment as a marker) and study design (for example, cluster‐randomised controlled trials versus individually randomised controlled trials) on the findings.
Results
Description of studies
Results of the search
We summarised the search results in detail in Figure 1. Briefly, we screened 2830 records of which we identified 26 as potentially eligible. Scrutinising the 26 full‐text articles resulted in four studies meeting our eligibility criteria; we excluded the remainder for the reasons displayed in the Characteristics of excluded studies table. We identified six ongoing studies, the available details of which are provided in the Characteristics of ongoing studies table. We entered non‐English abstracts into Google Translate to get a general idea of the study details. We did not need to obtain the full text for any non‐English study.
Flow diagram of search.
Included studies
We included four trials with 2894 children. All four studies were conducted in Malawi by a similar group of investigators (that is, same contact author and a number of co‐authors overlap). The articles Manary 2004 and Ndekha 2005 come from the same trial, but we included them as separate studies as they involved different children, namely with and without HIV. Three of the trials were cluster‐randomised. Using the effective sample size for the cluster‐randomised trials, the total number of children included in this review is 2594. Below is the list of included studies grouped according to the type of comparison. The Characteristics of included studies table provides further details.
Comparison 1: RUTF meeting total daily requirements versus standard diet (flour porridge)
-
Ciliberto 2005: quasi‐randomised cluster trial of 645 children with an effective sample size of 352 (HIV status not reported), aged between 10 and 60 months were assessed in Malawi. The RUTF was locally manufactured and the standard diet provided first in hospital and then at home.
-
Manary 2004: quasi‐randomised cluster trial of 186 HIV‐uninfected children (effective sample size 181) older than 12 months of age were assessed in Malawi. The RUTF was commercially produced.
-
Ndekha 2005: quasi‐randomised cluster trial of 65 HIV‐infected children (effective sample size 65) aged between 12 and 60 months were investigated in Malawi. The RUTF was commercially produced.
Comparison 2: RUTF supplement versus RUTF meeting total daily requirements
-
Manary 2004: quasi‐randomised cluster trial of 165 HIV‐uninfected children (effective sample size 161) older than 12 months of age were assessed in Malawi. The RUTF was commercially produced.
-
Ndekha 2005: quasi‐randomised cluster trial of 48 HIV‐infected children (effective sample size 48) aged between 12 and 60 months were investigated in Malawi. The RUTF was commercially produced.
Comparison 3: RUTF containing less milk powder versus standard RUTF
-
Oakley 2010: individually randomised trial of 1874 children (those known to be HIV‐infected were excluded from the trial) aged between 6 and 59 months in Malawi. RUTF was locally produced.
As the children of all included studies were severely malnourished, the interventions in three studies were given after the children had been stabilised with F75. Oakley 2010 did not report on stabilisation. All studies provided peanut‐based RUTF. The two studies that involved commercially produced RUTF used Plumpy'nut® (manufactured by Nutriset, Malaunay, France).
While we would have liked to report on the change (gain) in symptoms or signs from baseline for all continuous outcomes, this was not always possible as studies did not provide the information required to calculate this (for example, values at baseline and SD of change). In such cases, we assessed the difference in outcomes at the end of the intervention period. This is not an ideal approach given that the studies were generally small (effective sample sizes ranging from 93 to 1874; median 314) and only one was adequately randomised. The outcomes are solely dependent on anthropometrical measurements. Only one study sufficiently described how measurements were done allowing the reader to judge the quality (see the Characteristics of included studies table) (Oakley 2010).
Excluded studies
We excluded 22 studies. The most common reasons for exclusion were that the intervention was not a RUTF (seven studies) and that the study design was neither an RCT nor a quasi‐randomised trial (six studies). Furthermore, three studies only included moderately malnourished children; two were not home‐based; two compared commercially produced RUTF with non‐commercially produced RUTF; one looked at prevention (and not treatment), and one study examined the acceptability of RUTF rather than the efficacy or safety, or both. (See the Characteristics of excluded studies table.)
Risk of bias in included studies
We present our judgements regarding the risk of bias in each of the included studies in the Characteristics of included studies table. We found the risk of bias to be high for the three quasi‐randomised trials while the fourth trial had a low to moderate risk of bias. Two figures provide a graphical summary of the risk of bias assessments (Figure 2; Figure 3). We present additional information regarding the risk of bias in the three cluster‐randomised trials in Table 12.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
| Study ID | Recruitment bias | Baseline imbalance | Loss of clusters |
| Ciliberto 2005 | Inadequate Being a stepped wedge design, recruitment occurred after sites were assigned a specific treatment. All children eventually ended up with RUTF, although the time point at which conversion from standard care to RUTF took place was unknown The study authors recognised that recruitment bias was possible: "a source of bias might have been that a mother of a moderately malnourished child might have visited the NRU for screening when she heard that home‐based therapy was being offered" | Unclear Baseline characteristics per intervention arm were reported (significant difference in terms of WHZ), but similarities and differences between clusters were not mentioned | Adequate All randomised children were included in the analyses |
| Manary 2004 | Adequate Children were recruited after discharge days were allocated to a specific treatment. However, an independent doctor discharged the children unknowingly which discharge days matched which treatment. Therefore, the risk of recruitment bias was minimised | Unclear Baseline characteristics per intervention arm were reported, but similarities and differences between children discharged on different days were not mentioned | Adequate All randomised children were included in the analyses |
| Ndekha 2005 | Adequate Children were recruited after the weeks of discharge were allocated to a specific treatment. However, an independent doctor discharged the children unknowingly which discharge weeks matched which treatment. Therefore, the risk of recruitment bias was minimised | Unclear Baseline characteristics per intervention arm were reported, but similarities and differences between children discharged during different weeks were not mentioned | Adequate All randomised children were included in the analyses |
NRU: nutrition rehabilitation unit; RUTF: ready‐to‐use therapeutic food; WHZ: weight for height z score.
Allocation
Allocation refers to both the generation of the random allocation sequence and concealment of the allocation code.
We judged one included study to have a low risk of selection bias (Oakley 2010); the other three studies were quasi‐randomised and, therefore, we judged them to have a high risk of selection bias.
Blinding
We judged all four included studies to have a low risk of performance bias because the participants across groups received the same amount of contact time with study personnel. In terms of detection bias, only Oakley 2010 reported that outcome assessors were unaware of the intervention that the child received. However, it was not explained how blinding was ensured. In Ciliberto 2005, outcome assessors were not blinded, and in Manary 2004 and Ndekha 2005, blinding was not reported. As the majority of the primary and secondary outcomes were dependent on physical measurements by outcome assessors, we judged all included studies to have an unclear risk of detection bias.
Incomplete outcome data
We judged two studies to have a low risk of attrition bias as they did not have different losses to follow‐up in the two groups or large numbers of losses to follow‐up (Ciliberto 2005; Oakley 2010). We judged the other two studies to have a high risk of bias (Manary 2004; Ndekha 2005) due to different losses to follow‐up between the two treatment groups. It is worth noting that these two studies both had three arms: RUTF (total daily requirements), RUTF supplement and standard diet (flour porridge). In both these studies, the RUTF supplement arm had more than double the percentage attrition than the other two arms. We provided a summary of missing data in Table 11.
Selective reporting
For each included study, we searched for the protocol in the trial registries mentioned under Search methods for identification of studies and contacted the primary study authors asking whether their studies had been registered. Studies for which no protocol was available can at best be judged as having an unclear risk of bias with regards to selective reporting. This was the case in all studies except for Oakley 2010, which was registered. For the three included studies without a registered protocol, we assessed whether reports provided the prespecified study outcomes in the Methods section and judged those that did not specify their outcomes as having a high risk of bias. Ciliberto 2005 and Manary 2004 prespecified their outcomes and addressed them adequately; hence we judged them to have an unclear risk of bias. We judged studies that reported results for outcomes in addition to the prespecified outcomes, or those that prespecified outcomes but did not report such results, to have a high risk of bias (Ndekha 2005; Oakley 2010).
Other potential sources of bias
We judged Oakley 2010 to have a low risk of bias as the baseline characteristics in the two groups were similar, and the study was not funded by industry. In terms of baseline characteristics, we judged two studies to have an unclear risk of bias because there were significant differences in important baseline characteristics between the treatment and control groups (Ciliberto 2005; Ndekha 2005). We judged two studies in which a commercial RUTF manufacturer donated RUTF to have an unclear risk of bias (Manary 2004; Ndekha 2005). Furthermore, where there was risk of differential sharing of interventions by non‐participants (siblings and other family members) between comparison groups we considered this a further potential source of bias. We identified the studiesby Manary 2004 and Ndekha 2005 as having a risk of such sharing.
In cluster‐randomised trials, it is important to consider the unit of randomisation to avoid potential bias. We investigated this for the three cluster‐randomised trials (Manary 2004; Ciliberto 2005; Ndekha 2005). None of these studies calculated the effective sample size or accounted for clustering in their analyses. In addition, we assessed recruitment bias, cluster baseline imbalances and loss of clusters for these three studies (Table 12). We judged two studies (Manary 2004; Ndekha 2005) to have a low risk of recruitment bias because, although the children were recruited after the different clusters were allocated a specific intervention, an independent doctor discharged the children without knowing which discharge days matched which intervention. We judged one study (Ciliberto 2005) to have a high risk of recruitment bias because children were recruited after sites were assigned a specific intervention. In terms of baseline imbalances, we judged all three cluster‐randomised trials to have an unclear risk of bias because no relevant information was provided to assess this particular aspect. All clusters in the three cluster‐randomised trials were retained and, therefore, we judged them to have a low risk of bias regarding loss of clusters.
Effects of interventions
See: Summary of findings for the main comparison Ready‐to‐use therapeutic food (RUTF) compared to standard diet for children aged six months to five years with severe acute malnutrition; Summary of findings 2 Ready‐to‐use therapeutic food (RUTF) supplement compared to RUTF (total daily requirements) for children aged six months to five years with severe acute malnutrition; Summary of findings 3 Ready‐to‐use therapeutic food (RUTF) containing less milk powder compared to standard RUTF for children aged six months to five years with severe acute malnutrition
We were unable to analyse results by different age groups. There was no measurement of our secondary outcome, cognitive function and development, in any of the four trials. There was also no explicit measuring of allergic reactions as an adverse outcome.
Comparison 1: RUTF meeting total daily requirements versus standard diet (flour porridge)
Three trials with 896 children (effective sample size 598) evaluated the efficacy of RUTF versus flour porridge (Manary 2004; Ciliberto 2005; Ndekha 2005). All three trials assigned their participants in clusters, and all three were quasi‐randomised trials. While Ciliberto 2005 did not report on the participants' HIV status, Manary 2004 mentioned that they only included HIV‐uninfected children and Ndekha 2005 studied only HIV‐infected children. For the analyses below, we grouped HIV‐uninfected and HIV‐untested children and referred to them as HIV‐uninfected.
Primary outcomes
Recovery
Ciliberto 2005, Manary 2004 and Ndekha 2005 defined recovery as reaching a WHZ score > ‐2, having a WHZ score ≥ 0 and reaching 100% weight for height, respectively. More children recovered with RUTF than with standard diet irrespective of HIV status (RR 1.32; 95% CI 1.16 to 1.50; n = 599) (Analysis 1.1; Figure 4) and there was no significant heterogeneity detected between the studies (Chi2 = 1.59, degrees of freedom (df) = 2; P value = 0.45; I2 = 0%). After excluding Ndekha 2005, where all participants were confirmed to be HIV‐infected, the findings of the HIV‐uninfected subgroup remained similar (RR 1.32; 95% CI 1.10 to 1.58; n = 534) and there was no significant heterogeneity between the trials (Chi2 = 1.49, df = 1; P value = 0.22; I2 = 33%). In the HIV‐infected subgroup (Ndekha 2005), there was no difference between the RUTF and standard diet groups (RR 1.41; 95% CI 0.97 to 2.04; n = 65).
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.1 Recovery.
Relapse
Relapse was defined as admission to inpatient therapeutic care during the study period (Manary 2004; Ciliberto 2005; Ndekha 2005). In Manary 2004, results were reported as "died or relapsed". We contacted the study authors who sent us the data for "relapsed" and "died" separately. Due to heterogeneity (Chi2 = 5.58; df = 2; P value = 0.06; I2 = 64%), we did not pool results for these three trials (Analysis 1.2). This heterogeneity could not be explained by subgroup analysis as the two trials where children were HIV‐uninfected also had significant heterogeneity (Chi2 = 4.31; df = 1; P value = 0.04; I2 = 77%). The RRs for Ciliberto 2005 and Manary 2004 were 0.55 (95% CI 0.24 to 1.26; n = 352) and 0.05 (95% CI 0.00 to 0.74; n = 182), respectively. RUTF was not favoured compared to standard diet in HIV‐infected children (RR 0.10; 95% CI 0.01 to 1.70; n = 65) (Ndekha 2005).
Mortality
We detected no difference in mortality between the RUTF and standard diet groups (RR 0.97; 95% CI 0.46 to 2.05; n = 599) (Analysis 1.3; Figure 5), with no significant heterogeneity detected between the trials (Chi2 = 1.5; df = 2; P value = 0.47; I2 = 0%) (Manary 2004; Ciliberto 2005; Ndekha 2005). Subgroup results for HIV‐uninfected children (RR 0.78; 95% CI 0.32 to 1.88; n = 534) (Manary 2004; Ciliberto 2005) (no significant heterogeneity detected between the trials, Chi2 = 0.66; df = 1; P value = 0.42; I2 = 0%) did not differ from subgroup results for children diagnosed with HIV (Ndekha 2005) (RR 1.69; 95% CI 0.42 to 6.85; n = 65).
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.3 Mortality.
Secondary outcomes
Weight gain
Manary 2004, Ciliberto 2005 and Ndekha 2005 reported weight gain measured during the first four weeks of the intervention period. We obtained data on weight gain from the contact author of Manary 2004 and pooled the results of the three trials. We found that the RUTF group gained more weight when compared to the standard diet group (MD 1.47 g/kg/day; 95% CI 0.49 to 2.45; n = 595) (Analysis 1.4; Figure 6) and there was no significant heterogeneity detected between the trials (Chi2 = 2.92; df = 2; P value = 0.23; I2 = 32%). When separating these results in subgroup analyses by HIV status, there was also a difference favouring RUTF in HIV‐uninfected children (MD 1.79 g/kg/day; 95% CI 0.65 to 2.93; n = 530) (Manary 2004; Ciliberto 2005) and there was no significant heterogeneity detected between the trials (Chi2 = 1.28; df = 1; P value = 0.26; I2 = 22%). However, in the subgroup of HIV‐infected children, RUTF was not favoured above standard diet (MD 0.80 g/kg/day; 95% CI ‐0.64 to 2.24; n = 65) (Ndekha 2005).
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.4 Weight gain (g/kg/day).
Time to recovery
Ndekha 2005 measured time to recovery and reported results in median days. HIV‐infected children in the RUTF group (n = 20) recovered within a median of 71 days (interquartile range 42 to 125) compared to 85 days (interquartile range 46 to 239) for the standard diet group (n = 45). The authors did not report a P value or significance for the difference in time to recovery between these two groups. Manary 2004 also reported time to recovery but results were displayed only in a graph from which accurate information could not be obtained for further analysis. We contacted the trial authors, who provided us with the necessary information. No difference in time to recovery was detected in HIV‐uninfected children between the treatment and standard diet groups (MD ‐7.0 days; 95% CI ‐15.89 to 1.89; n = 136) (Analysis 1.5).
Mid‐upper arm circumference (MUAC) gain
Three trials measured MUAC during the first four weeks of the intervention period. Data in Manary 2004 were reported only in graph form and we obtained the actual values from the contact author. We found that children in the RUTF group had a higher MUAC gain compared to children with standard diet (MD 0.13 mm/day; 95% CI 0.04 to 0.21; n = 570) (Analysis 1.6) and there was no significant heterogeneity detected between trials in all children (Chi2 = 2.3; df = 2; P value = 0.32; I2 = 13%). After excluding results for children with HIV, a subgroup analysis of the HIV‐uninfected subgroup also found a higher MUAC gain with RUTF compared with standard diet (MD 0.15 mm/day; 95% CI 0.07 to 0.24; n = 505) (Manary 2004; Ciliberto 2005) and there was no significant heterogeneity between the trials (Chi2 = 0.05; df = 1; P value = 0.82; I2 = 0%). In the HIV‐infected subgroup, however, we detected no difference between RUTF and standard diet (MD ‐0.04 mm/day; 95% CI ‐0.28 to 0.20; n = 65) (Ndekha 2005).
Weight for height z score (WHZ)
HIV‐uninfected children who recovered and were discharged from the trial were followed up for six months (Manary 2004). We obtained results from the contact author. There was no difference detected in WHZ between the RUTF and standard diet group (MD 0.19; 95% ‐0.22 to 0.60; n = 99; Analysis 1.7). We could not report on WHZ gain because the SD of change or the exact P value of the difference between the change in each group were not reported. Since baseline values for WHZ across the different groups were similar, results for WHZ gain would not have been different from the WHZ that we reported above.
Adverse outcomes
The numbers of days of diarrhoea per group were measured during the first two weeks of the treatment period (Ciliberto 2005). Children who received RUTF had a similar frequency of diarrhoea to those on standard diet (MD ‐0.6; 95% CI ‐1.30 to 0.10; n = 352) (Analysis 1.8). Ndekha 2005 measured the "prevalence of diarrhoea" (days of diarrhoea divided by the "total days" during the first two weeks of the treatment period) in HIV‐infected children. Children in the RUTF group had diarrhoea on 19 out of the 304 evaluated days compared to 57 out of 687 evaluated days in children in the standard diet group. Since the corresponding numbers of participants were not reported, a treatment effect could not be calculated.
Comparison 2: RUTF supplement versus RUTF meeting total daily requirements
Two quasi‐randomised cluster trials that used systematic sequence generation methods had the following three arms: (1) RUTF meeting total daily nutritional requirements, (2) the same RUTF given supplementary to children's habitual diet and (3) standard diet (maize/soy flour) (Manary 2004; Ndekha 2005). Below we compared the RUTF (total daily requirements) with RUTF supplement. It is also important to note that Manary 2004 included only HIV‐uninfected children while Ndekha 2005 only assessed HIV‐infected children.
Primary outcomes
Recovery
Children who received the supplement were less likely to recover than those who received RUTF (RR 0.71; 95% CI 0.60 to 0.84; n = 210) (Manary 2004; Ndekha 2005) (Analysis 2.1; Figure 7) and there was no significant heterogeneity detected between the trials (Chi2 = 0.37; df = 1; P value = 0.54; I2 = 0%). We made the same conclusion when considering HIV‐uninfected and HIV‐infected children in subgroups separately (RR 0.72; 95% CI 0.60 to 0.87; n = 162 for Manary 2004 and RR 0.62; 95% CI 0.39 to 0.99; n = 48 for Ndekha 2005).
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.1 Recovery.
Relapse
Manary 2004 and Ndekha 2005 measured relapse (admission to hospital) during the intervention period. Pooled results indicated that children were more likely to have relapsed in the supplement group (13 out of 122 children) when compared to the RUTF group where none of the 88 children relapsed (RR 8.95; 95% CI 1.18 to 67.77; n = 210) (Analysis 2.2; Figure 8). There was no significant heterogeneity between the trials (Chi2 = 0.3; df = 1; P value = 0.58; I2 = 0%). However, when separating the results into subgroups based on HIV status, there was no difference detected between the two groups (RR 15.25; 95% CI 0.91 to 255.9; n = 162; (Manary 2004) and RR 5.07; 95% CI 0.28 to 93.0; n = 48 (Ndekha 2005)).
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.2 Relapse.
Mortality
When comparing the supplement group with the RUTF group (Manary 2004; Ndekha 2005), we detected no difference in mortality (RR 0.73; 95% CI 0.25 to 2.18; n = 210) (Analysis 2.3; Figure 9) and there was no significant heterogeneity between the trials (Chi2 = 0.36; df = 1; P value = 0.55; I2 = 0%). Similarly, we detected no difference in mortality when assessing results for the HIV‐uninfected (Manary 2004) and HIV‐infected (Ndekha 2005) subgroups separately (RR 0.48; 95% CI 0.08 to 2.81; n = 162 and RR 0.95; 95% CI 0.24 to 3.80; n = 48, respectively).
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.3 Mortality.
Secondary outcomes
Weight gain
We did not pool results for RUTF as a supplement versus RUTF (total daily requirements) because of substantial heterogeneity (Chi2 = 4.26; df = 1; P value = 0.04; I2 = 76.5%) (Analysis 2.4) (Manary 2004; Ndekha 2005). Both of these trials measured weight gain during the first four weeks of the intervention period. Manary 2004 found HIV‐uninfected children in the supplement group gained less weight than children who received RUTF (MD ‐2.10 g/kg/day; 95% CI ‐3.08 to ‐1.12; n = 158). For HIV‐infected children, we detected no difference in weight gain between the two groups (MD ‐0.10 g/kg/day; 95% CI ‐1.73 to 1.53; n = 48) (Ndekha 2005).
Time to recovery
HIV‐uninfected children who received RUTF supplementary to their habitual diet recovered faster than those from the RUTF (total daily requirements) group (MD 10.0 days; 95% CI 0.87 to 19.13; n = 116) (Analysis 2.5) (Manary 2004). The HIV‐infected children in the supplement group required a median of 115 days (interquartile range 59 to 195) to reach 100% weight for height while those in the RUTF group recovered within a median of 71 days (interquartile range 42 to 125) (WHO/National Center for Health Statistics (NCHS) standard) (P value not reported in the article) (Ndekha 2005).
Mid‐upper arm circumference (MUAC) gain
We detected no difference in MUAC gain during the first four weeks of the intervention period between the two groups (MD ‐0.11 mm/day; 95% CI ‐0.22 to 0.01; n = 173) (Analysis 2.6) (Manary 2004; Ndekha 2005) and no significant heterogeneity was detected between the trials (Chi2 = 1.35; df = 1; P value = 0.25; I2 = 26%). When separating results for HIV‐infected and HIV‐uninfected subgroups, children in the supplement group fared worse than children in the RUTF group in HIV‐uninfected children (MD ‐0.15 mm/day; 95% CI ‐0.27 to ‐0.03; n = 125) (Manary 2004) while in HIV‐infected children there was no difference detected (MD ‐0.03 mm/day; 95% CI ‐0.20 to 0.14; n = 48) (Ndekha 2005).
Weight for height z scores (WHZ)
HIV‐uninfected children who recovered and were discharged from the trial were followed up for six months (Manary 2004). Results were obtained from the study author, which indicated that there was no difference in WHZ between children who received RUTF as a supplement and those who received RUTF (total daily requirements) (MD ‐0.10; 95% CI ‐0.56 to 0.36; n = 73). There were no baseline WHZ differences between the groups. Therefore, the differences in WHZ gain between the groups were not compared and would not have made a significant impact on the findings.
Adverse outcomes
Ndekha 2005 measured the "prevalence of diarrhoea" (days of diarrhoea divided by the "total days" during the first two weeks of the intervention period) in HIV‐infected children. Study authors reported that children in the supplement group had diarrhoea on 38 out of the 432 days while children in the RUTF group had diarrhoea on 19 out of the 304 evaluation days. We were unsure about the meaning of "total days" as the figures did not correspond to the total number of days of each participant for each group. Since the corresponding numbers of participants were not reported, no treatment effect could be calculated.
Comparison 3: RUTF containing less milk powder versus standard RUTF
In a comparison of standard RUTF (25% milk powder) versus a formula of RUTF that contains less milk powder (10%), we found one individually randomised trial with 1874 HIV‐uninfected children between 6 and 59 months of age in Malawi (Oakley 2010).
Primary outcomes
Recovery
Oakley 2010 defined recovery as having a WHZ greater than ‐2 without oedema within a maximum of eight weeks. We detected no difference in recovery between the two groups who received different formulations of RUTF (RR 0.97; 95% CI 0.93 to 1.01; n = 1874) (Analysis 3.1).
Relapse
Oakley 2010 defined relapse as the number of children who remained wasted plus the number of children who were referred to inpatient care. More children in the 10% milk RUTF group relapsed than children in the 25% milk RUTF group (RR 1.33; 95% CI 1.03 to 1.72; n = 1874) (Analysis 3.2).
Mortality
We found that the group that received the 10% milk RUTF had a similar number of deaths compared with the 25% milk RUTF group (RR 0.90; 95% CI 0.55 to 1.45; n = 1874) (Analysis 3.3) (Oakley 2010).
Secondary outcomes
Weight gain
Oakley 2010 measured weight gain during the intervention period. Children in the 10% milk RUTF group gained less weight than children in the 25% milk RUTF group (MD ‐0.50 g/kg/day; 95% CI ‐0.75 to ‐0.25; n = 1874) (Analysis 3.4).
Mid‐upper arm circumference (MUAC) gain
Oakley 2010 measured MUAC. The time point was not reported. Children who received the 10% milk RUTF had less MUAC gain than children who received the 25% milk RUTF (MD ‐0.04 mm/day; 95% CI ‐0.06 to ‐0.02; n = 1874) (Analysis 3.5).
Weight for height z score (WHZ)
When comparing the end values between the two groups who received different RUTF formulations, we detected no difference in WHZ (MD 0.00; 95% CI ‐0.10 to 0.10; n = 1874) (Analysis 3.6) (Oakley 2010).
Weight for age z score (WAZ)
When comparing the end values between the 10% milk RUTF and 25% milk RUTF groups, we detected no difference in WAZ (MD ‐0.10; 95% CI ‐0.21 to 0.01; n = 1874) (Analysis 3.7) (Oakley 2010).
Height for age z score (HAZ)
When comparing the end values between the two groups who received different RUTF formulations, we detected no difference in HAZ (MD ‐0.10; 95% CI ‐0.24 to 0.04; n = 1874) (Analysis 3.8) (Oakley 2010).
Adverse outcomes
Oakley 2010 did not measure adverse outcomes.
Discussion
Summary of main results
Our review aimed to assess the effects of home‐based RUTF on relapse, mortality and weight gain in children with SAM. We found four eligible studies, of which three were quasi‐randomised cluster trials that addressed three different comparisons. The total effective sample size was 2594 children. We found the risk of bias to be high for the three quasi‐randomised trials, while the fourth trial had a low to moderate risk of bias. One of the studies included HIV‐infected children, and all relevant meta‐analyses were subgrouped according to HIV status. The number of studies and sample size per subgroup was too small to justify reporting the summarised results separately. We have displayed the most important findings in summary of findings Table for the main comparison, summary of findings Table 2 and summary of findings Table 3 and summarised our findings narratively below.
RUTF meeting total daily requirements versus standard diet (flour porridge)
When comparing RUTF versus standard diet (flour porridge) we found three quasi‐randomised cluster trials (n = 599). RUTF may improve recovery slightly (RR 1.32; 95% CI 1.16 to 1.50; low‐quality evidence), but we do not know whether RUTF improves relapse, mortality or weight gain (very low quality evidence).
RUTF supplement versus RUTF meeting total daily requirements
When comparing RUTF supplement versus RUTF that meets total daily nutritional requirements we found two quasi‐randomised cluster trials (n = 210). For recovery, relapse, mortality and weight gain the quality of evidence was very low; therefore, the effects of RUTF are unknown.
RUTF containing less milk powder versus standard RUTF
When comparing a cheaper RUTF containing less milk powder (10%) versus standard RUTF (25% milk powder), we found one trial that randomised 1874 children. For recovery, there probably was little or no difference between the groups (RR 0.97; 95% CI 0.93 to 1.01; moderate‐quality evidence). RUTF containing less milk powder may lead to slightly more children relapsing (RR 1.33; 95% CI 1.03 to 1.72; low‐quality evidence) and to less weight gain (MD ‐0.5 g/kg/day; 95% CI ‐0.75 to ‐0.25; low‐quality evidence) than standard RUTF. We do not know whether the cheaper RUTF improves mortality (very low quality evidence).
Overall completeness and applicability of evidence
In this review, we sought to evaluate the best evidence regarding the efficacy and safety of RUTF as home‐based treatment. Should RUTF be found to be more efficacious than the standard diet, then, from a health systems perspective, it would be important to know whether the cheaper RUTF regimen (RUTF as a supplement) and formulation (reduced milk powder content) could achieve similar or better health outcomes. From a nutritional perspective, it is important that the children's carers sustain and improve cultural‐specific dietary habits instead of relying solely on provided medical nutritional therapy. The abovementioned issues informed the three comparisons investigated in this systematic review. Although a number of trials have been conducted with RUTF (for example, see the Characteristics of excluded studies table), a limited number of randomised controlled trials investigated the comparisons we identified as being most important for SAM children.
The studies included in this review have a number of limitations in respect of external validity. For example, each of the studies assessed a wide age range and did not allow for exploration of possible differences in effect between younger and older children. Furthermore, comorbidities such as HIV, which may have considerable impacts on growth and immunity, are not sufficiently addressed in the current body of evidence. There is a lack of information on participants' total daily energy intake per group as well as the likelihood of sharing RUTF and standard diet with the family and whether something was done to prevent differential sharing. In addition, the issue of allergies was not sufficiently addressed. In all four included studies, children were exposed to peanuts and soy, which are both known allergens. However, no study tested for soy allergies, and only two studies tested for peanut allergies (Manary 2004; Ndekha 2005).
No included study followed up children for more than six months and, therefore, we could not evaluate potential long‐term growth and developmental differences. Specifically, no study assessed cognitive function and development, which is important for future school performance. Furthermore, studies employed different definitions for outcomes such as recovery (different reference standards and cut‐off points) and anthropometrical measurements (assessed at different time points). One of the theoretical benefits of RUTF is a low water availability (thus less likely to become contaminated with microorganisms), which should lead to fewer episodes of diarrhoea. However, diarrhoea was not a primary outcome in any of the included studies, despite the fact that diarrhoea is one of the biggest causes of mortality in young children. More emphasis must be placed on this in future RUTF research. Overall, we consider that the current body of evidence for the three comparisons addressed in this review lacks important information to allow evaluation of applicability in different types of children and settings.
Quality of the evidence
The reporting of trials included in this review was generally poor, thus necessitating contact with the trial authors. We are grateful to the authors for providing data requested for our analysis. With our risk of bias assessment, we identified the following to be of high concern: selection bias (as three of the four included studies were quasi‐randomised trials), attrition bias and reporting bias. The few included studies were not suitable for sensitivity analyses. As shown by our GRADE assessments (summary of findings Table for the main comparison; summary of findings Table 2; summary of findings Table 3), the quality of evidence varied between moderate and very low, which means that future research is likely to impact on the findings. Therefore, our confidence in the findings is limited.
One key methodological limitation of the included studies is the lack of a definition for SAM. Only Oakley 2010 explicitly defined SAM. Baseline characteristics were taken upon enrolment (after the stabilisation period) while diagnosis of SAM was made before stabilisation. Therefore, improvement in nutritional status during the stabilisation period is not reflected in the baseline characteristics reported in the included articles, which raises concern about whether the children were severely malnourished at the start of the trials.
Potential biases in the review process
It is unlikely that we have missed any relevant trials since, apart from our electronic search without any language restriction, we also contacted the corresponding authors of the included studies (and some of the excluded studies) as well as professionals working in the field.
We were unable to assess the likelihood of publication bias formally due to the small number of studies per comparison.
Agreements and disagreements with other studies or reviews
To our knowledge, this is the first systematic review that specifically compared home‐based RUTF with standard diet for the treatment of SAM children. However, we are aware of another review where the "efficacy and safety of home‐based management of SAM using 'therapeutic nutrition products' or ready to use therapeutic foods and efficacy of these products in comparison with F100 and home‐based diet" were assessed (Gera 2010). We evaluated the methodological quality of Gera 2010 with the validated AMSTAR tool (Shea 2007), and presented our findings in Table 13. Although Gera 2010 was published as a systematic review, the methods followed did not meet all of the requirements.
| Yes/No/Can't answer/Not applicable | ||
| Criteria | Our RUTF review | |
| 1. Was an 'a priori' design provided? ‐ The research question and inclusion criteria should be established before the conduct of the review | Yes | Yes |
| 2. Was there duplicate study selection and data extraction? | No | Yes |
| 3. Was a comprehensive literature search performed? | Incomplete ‐ CENTRAL and MEDLINE were searched with 4 different search terms (no search strings) on 20 April 2010. This electronic database search was not supplemented by screening reference lists and contacting researchers/clinicians in the field | Yes |
| 4. Was the status of publication (i.e. grey literature) used as an inclusion criterion? | Can't answer, nothing about publication and language was reported | No |
| 5. Was a list of studies (included and excluded) provided? ‐ A list of included and excluded studies should be provided | No ‐ A list of excluded studies was not reported | Yes |
| 6. Were the characteristics of the included studies provided? | No ‐ A list of included studies was provided but it did not contain all relevant information (e.g. characteristics of participants, interventions, control) | Yes |
| 7. Was the scientific quality of the included studies assessed and documented? | No | Yes |
| 8. Was the scientific quality of the included studies used appropriately in formulating conclusion? | No | Yes |
| 9. Were the methods used to combine the | No ‐ 2 meta‐analyses were performed, 1 that was accompanied by a I2 value of 89.8% and 1 did not report an I2 value. The term 'heterogeneity' did not appear in the article. In addition, the software used to conduct the meta‐analyses was not reported and nothing mentioned about the effects model (i.e. fixed or random) used | Yes |
| 10. Was the likelihood of publication bias assessed? | No ‐ Publication bias was not mentioned in the article | Not applicable ‐ We set out in our protocol to assess the likelihood of publication bias, but since we had too few studies per comparison we could not draw a funnel plot |
| 11. Was the conflict of interest stated? | Yes ‐ The author declared that she had no competing interests and she stated that there was no funding | Yes |
Gera 2010 included two reviews, seven "controlled trials", seven cohort studies and two consensus statements. The outcomes assessed were recovery rate (as defined by the study authors), weight gain (g/kg/day), relapse, mortality and morbidities (for example, diarrhoea, malaria and respiratory infections). Of the seven "controlled trials", we have included two in our review (Manary 2004; Ciliberto 2005). We did not include the remaining five trials because one was facility‐based (Diop 2003), three did not have an eligible control group (Diop 2004; Sandige 2004; Gabouland 2007), and one trial included Spirulina® and not RUTF (Simpore 2006). Gera 2010 did not perform a meta‐analysis of trials on any of the comparisons that we evaluated in our review but pooled four cohort studies that assessed the effect of home‐based RUTF on weight gain in SAM children and found a "mean weight gain" (type of effect size not specified) of 3.2 g/kg/day (95% CI 3.06 to 3.34; I2 = 89.8%) (software and effects model not reported). In our review, we also found that children who received RUTF as opposed to standard diet gained more weight. However, our MD is smaller (MD 1.47 g/kg/day; 95% CI 0.49 to 2.45; random‐effects analysis, heterogeneity Chi2 = 2.92; I2 = 32%; Analysis 1.4).
Ashworth 2006 and Bhutta 2008 evaluated the efficacy of interventions for malnutrition in young children, which included RUTF. While the authors did not comment on the effectiveness of RUTF versus standard diet, they did report that RUTF could be used in home‐based settings.
Another Cochrane systematic review that will evaluate nutritional therapy for malnutrition is underway. Lazzerini 2012 is evaluating the "safety and effectiveness of different types of foods for children with moderate acute malnutrition (MAM) in low‐ and middle‐income countries". RUTF will be included in this review, and when completed, will complement our systematic review relating to SAM children.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.1 Recovery.
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.3 Mortality.
Forest plot of comparison: 1 RUTF versus standard diet, outcome: 1.4 Weight gain (g/kg/day).
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.1 Recovery.
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.2 Relapse.
Forest plot of comparison: 2 RUTF (total daily requirements) versus RUTF supplement, outcome: 2.3 Mortality.
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 1 Recovery.
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 2 Relapse.
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 3 Mortality.
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 4 Weight gain (g/kg/day).
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 5 Time to recovery for HIV‐uninfected children (days).
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 6 Mid‐upper arm circumference gain (mm/day).
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 7 Weight for height z score at follow‐up in HIV‐uninfected children.
Comparison 1 Ready‐to‐use therapeutic food (RUTF) versus standard diet, Outcome 8 Days of diarrhoea during the intervention period.
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 1 Recovery.
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 2 Relapse.
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 3 Mortality.
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 4 Weight gain (g/kg/day).
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 5 Time to recovery for HIV‐uninfected children (days).
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 6 Mid‐upper arm circumference gain (mm/day).
Comparison 2 Ready‐to‐use therapeutic food (RUTF) supplement versus RUTF (total daily requirements), Outcome 7 Weight for height z score at follow‐up for HIV‐uninfected children.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 1 Recovery.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 2 Relapse.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 3 Mortality.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 4 Weight gain (g/kg/day).
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 5 Mid‐upper arm circumference gain (mm/day).
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 6 Weight for height z score.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 7 Weight for age z score.
Comparison 3 Ready‐to‐use therapeutic food (RUTF) with reduced milk powder content versus RUTF with recommended milk powder content, Outcome 8 Height for age z score.
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Standard diet | RUTF | |||||
| Recovery | 597 per 1000 | 788 per 1000 | RR 1.32 | 599 | ⊕⊕⊝⊝ | ‐ |
| Relapse | See comment | See comment | Not estimable | 599 | ⊕⊝⊝⊝ | 2 studies found a large effect with RUTF, 1 study did not detect an effect5 |
| Mortality | 54 per 1000 | 53 per 1000 | RR 0.97 | 599 | ⊕⊝⊝⊝ | ‐ |
| Weight gain | ‐ | The mean weight gain in the intervention groups was | ‐ | 595 | ⊕⊝⊝⊝ | ‐ |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The three studies had different definitions for recovery. Ciliberto 2005: reaching WHZ > ‐2; Manary 2004: reaching WHZ ≥ 0; Ndekha 2005: reaching 100% weight for height. | ||||||
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| RUTF (total daily requirements) | RUTF supplement | |||||
| Recovery | 830 per 1000 | 589 per 1000 | RR 0.71 | 210 | ⊕⊝⊝⊝ | ‐ |
| Relapse | 0 per 1000 | 0 per 1000 | RR 8.95 | 210 | ⊕⊝⊝⊝ | ‐ |
| Mortality | 68 per 1000 | 50 per 1000 | RR 0.73 | 210 | ⊕⊝⊝⊝ | ‐ |
| Weight gain: HIV‐uninfected children | ‐ | The mean weight gain: HIV‐uninfected children in the intervention groups was | ‐ | 158 | ⊕⊝⊝⊝ | Weight gain in g/kg/day for a 6‐kg child translates to a clinically important difference of MD 378 g/month (95% CI 201.6 to 554.4) |
| Weight gain: HIV‐infected children | ‐ | The mean weight gain: HIV‐infected children in the intervention groups was | ‐ | 48 | ⊕⊝⊝⊝ | Weight in g/kg/day for a 6‐kg child translates to a clinically important difference of either a loss of 311.4 g/month or a gain of 275.4 g/month |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The 2 studies used different definitions for recovery. Manary 2004: reaching WHZ ≥ 0; Ndekha 100% weight for height. | ||||||
| Patient or population: children aged 6 months to 5 years with severe acute malnutrition | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Standard RUTF | RUTF containing less milk powder | |||||
| Recovery | 836 per 1000 | 811 per 1000 | RR 0.97 | 1874 | ⊕⊕⊕⊝ | ‐ |
| Relapse | 98 per 1000 | 131 per 1000 | RR 1.33 | 1874 | ⊕⊕⊝⊝ | ‐ |
| Mortality | 36 per 1000 | 32 per 1000 | RR 0.90 | 1874 | ⊕⊝⊝⊝ | ‐ |
| Weight gain | ‐ | The mean weight gain in the intervention groups was | ‐ | 1874 | ⊕⊕⊝⊝ | Weight in g/kg/day for a 6‐kg child translates to a clinically unimportant difference of a loss of MD 90 (95% CI 135 to 45) g/month |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 The intervention period was up to 8 weeks, but outcome data for children who recovered, relapsed or died before 8 weeks was captured until the time the child was finished with the study. | ||||||
| Severe acute malnutrition with complications | Severe acute malnutrition without complications |
| Bilateral pitting oedema grade 3*** (severe oedema) OR | MUAC < 110† mm OR |
| MUAC < 110† mm and bilateral pitting oedema grades 1* or 2** (marasmic kwashiorkor) OR | Bilateral pitting oedema grades 1* or 2** with MUAC ≥ 110† mm AND
|
| MUAC < 110† mm or bilateral pitting oedema grades 1* or 2** AND 1 of the following:
| ‐ |
| ↓ | ↓ |
| Inpatient care IMCI/WHO protocol | Outpatient therapeutic care protocols |
| IMCI: Integrated Management of Childhood Illness; MUAC: mid‐upper arm circumference; UNICEF: United Nations Children's Fund; WHO: World Health Organization. *Grade 1 = mild oedema on both feet or both ankles. **Grade 2 = moderate oedema on both feet, and on lower legs, hands or lower arms. ***Grade 3 = severe generalised oedema affecting feet, legs, hands, arms and face. †The WHO and UNICEF now recommend that the cut‐off value for the MUAC for severe acute malnutrition should be increased to 115 mm (WHO and UNICEF 2009). The adoption of this higher cut‐off value will sharply increase the caseloads, which may influence the cost of nutrition programmes greatly (WHO and UNICEF 2009). However, detecting more children earlier as severely malnourished will lead to a shorter period needed to treat them, which may bring down the cost per child (WHO and UNICEF 2009). ¤IMCI criteria: 60 respirations/min children age < 2 months; 50 respirations/min for age 2‐12 months; 40 respirations/min for age 1‐5 years; 30 respirations/min for age > 5 years. | |
| Moisture content | 2.5% maximum |
| Energy | 520‐550 kCal/100 g |
| Protein | 101‐2% total energy |
| Lipids | 45‐60% total energy |
| Sodium | 290 mg/100 g maximum |
| Potassium | 1110‐1400 mg/100 g |
| Calcium | 300‐600 mg/100 g |
| Phosphorus (excluding phytate) | 300‐600 mg/100 g |
| Magnesium | 80‐140 mg/100 g |
| Iron | 10‐14 mg/100 g |
| Zinc | 11‐14 mg/100 g |
| Copper | 1.4‐1.8 mg/100 g |
| Selenium | 20‐40 μg |
| Iodine | 70‐140 μg/100 g |
| Vitamin A | 0.8‐1.1 mg/100 g |
| Vitamin D | 15‐20 μg/100 g |
| Vitamin E | 20 mg/100 g minimum |
| Vitamin K | 15 to 30 μg/100 g |
| Vitamin B1 | 0.5 mg/100 g minimum |
| Vitamin B2 | 1.6 mg/100 g minimum |
| Vitamin C | 50 mg/100 g minimum |
| Vitamin B6 | 0.6 mg/100 g minimum |
| Vitamin B12 | 1.6 μg/100 g minimum |
| Folic acid | 200 μg/100 g minimum |
| Niacin | 5 mg/100 g minimum |
| Pantothenic acid | 3 mg/100 g minimum |
| Biotin | 60 μg/100 g minimum |
| n‐6 fatty acids | 3‐10% of total energy |
| n‐3 fatty acids | 0.3‐2.5% of total energy |
| Ingredient | % weight |
| Full‐fat milk | 30 |
| Sugar | 28 |
| Vegetable oil | 15 |
| Peanut butter* | 25 |
| Mineral‐vitamin mix | 1.6 |
| *Strict quality control is essential. | |
| Ingredient | Quantities (%) |
| Roasted rice flour | 20 |
| Soyamin 90 | 29 |
| Roasted sesame seed paste | 8 |
| Sunflower oil | 19.4 |
| Icing sugar | 22 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
| RUTF: ready‐to‐use therapeutic food. | |
| Ingredient | Quantities (%) |
| Roasted pearl barley flour | 15 |
| Soyamin 90 | 9 |
| Roasted sesame seed paste | 27 |
| Sunflower oil | 24 |
| Icing sugar | 23.4 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
| RUTF: ready‐to‐use therapeutic food. | |
| Ingredient | Quantities (%) |
| Roasted maize flour | 33.4 |
| Roasted sesame seed paste | 27 |
| Roasted chick pea flour | 25 |
| Sunflower oil | 12 |
| Icing sugar | 15 |
| Vitamin and mineral premix (CMV therapeutique, Nutriset) | 1.6 |
| RUTF: ready‐to‐use therapeutic food. | |
| Nutrient | Unit | RUTF‐1 (100 g) | Energy (%) | RUTF‐2 (100 g) | Energy (%) | RUTF‐3 (100 g) | Energy (%) | Plumpy'nut®* (100 g) | Energy (%) |
| Energy** | kCal | 551 | ‐ | 567 | ‐ | 512 | ‐ | 530 | ‐ |
| Energy | kJ | 2307 | ‐ | 2373 | ‐ | 2142 | ‐ | 2218 | ‐ |
| Protein | g | 13.8 | 10 | 14.1 | 10 | 13.4 | 11 | 14.5 | 11 |
| Carbohydrate*** | g | 43 | 31 | 39.9 | 28 | 50.2 | 39 | 43 | 32 |
| Fat | g | 36 | 59 | 39 | 62 | 28.6 | 50 | 33.5 | 57 |
| Ash | g | 43 | ‐ | 3.9 | ‐ | 4.9 | ‐ | 4 | ‐ |
| Moisture | g | 2.9 | ‐ | 3.1 | ‐ | 2.9 | ‐ | < 5 | ‐ |
| *Protein and fat are reported to contribute 11% and 57% in energy input, respectively. Total energy is reported to be 530 kcal/100 g and moisture < 5%. **The energy has been calculated using Atwater factors. ***Carbohydrate is by difference assuming protein to be nitrogen multiplied by 6.25. | |||||||||
| Mineral | RUTF‐1 (mg/kg) | RUTF‐2 (mg/kg) | RUTF‐3 (mg/kg) | Plumpy'nut®(mg/kg) |
| Cu | 2.1 | 2.1 | 1.8 | 1.7 |
| Zn | 10.9 | 10.9 | 10.2 | 13 |
| Ca | 338.1 | 338.1 | 209.8 | 310 |
| Na | 256.5 | 256.5 | 189.9 | < 290 |
| Mg | 118.4 | 118.4 | 119.1 | 86 |
| Fe | 5.6 | 5.6 | 4.4 | 12.45 |
| RUTF recipe | Water activity |
| RUTF‐1 | 0.290 |
| RUTF‐2 | 0.279 |
| RUTF‐3 | 0.260 |
| Plumpy'nut® | 0.241 |
| Independent additional criteria |
|
|
| ↓ | ↓ | |
| Type of therapeutic feeding | Facility‐based | Community‐based |
| Intervention |
|
|
| Discharge criteria (transition criteria from facility to community‐based care) |
|
|
| RUTF: ready‐to‐use therapeutic food; UNICEF: United Nations Children's Fund; WHO: World Health Organization. *Children who eat at least 75% of their calculated RUTF ration for the day. | ||
| Study ID | Participants recruited (n) | Pre‐randomisation attrition (n) | Immediately post‐randomisation attrition (n) | Drop‐outs during the intervention period (n) |
| Ciliberto 2005 | 1178 (includes moderate and severe malnourished children) | 0 | 41 (reasons not reported) | 72 (reasons not reported) |
| Manary 2004 | 452 | 77 refused 93 HIV positive | 0 | 47 were "dropouts" (no reasons reported) and 37 died (no reasons reported) |
| Ndekha 2005 | 93 | 0 | 0 | 11 died (no reasons reported) and 17 drop‐outs (no reasons reported) |
| Oakley 2010 | 1961 | 87 (reasons not reported) | 0 | 51 were lost to follow‐up (no reasons reported other than "those lost were more likely to be younger and marasmic") |
| Study ID | Recruitment bias | Baseline imbalance | Loss of clusters |
| Ciliberto 2005 | Inadequate Being a stepped wedge design, recruitment occurred after sites were assigned a specific treatment. All children eventually ended up with RUTF, although the time point at which conversion from standard care to RUTF took place was unknown The study authors recognised that recruitment bias was possible: "a source of bias might have been that a mother of a moderately malnourished child might have visited the NRU for screening when she heard that home‐based therapy was being offered" | Unclear Baseline characteristics per intervention arm were reported (significant difference in terms of WHZ), but similarities and differences between clusters were not mentioned | Adequate All randomised children were included in the analyses |
| Manary 2004 | Adequate Children were recruited after discharge days were allocated to a specific treatment. However, an independent doctor discharged the children unknowingly which discharge days matched which treatment. Therefore, the risk of recruitment bias was minimised | Unclear Baseline characteristics per intervention arm were reported, but similarities and differences between children discharged on different days were not mentioned | Adequate All randomised children were included in the analyses |
| Ndekha 2005 | Adequate Children were recruited after the weeks of discharge were allocated to a specific treatment. However, an independent doctor discharged the children unknowingly which discharge weeks matched which treatment. Therefore, the risk of recruitment bias was minimised | Unclear Baseline characteristics per intervention arm were reported, but similarities and differences between children discharged during different weeks were not mentioned | Adequate All randomised children were included in the analyses |
| NRU: nutrition rehabilitation unit; RUTF: ready‐to‐use therapeutic food; WHZ: weight for height z score. | |||
| Yes/No/Can't answer/Not applicable | ||
| Criteria | Our RUTF review | |
| 1. Was an 'a priori' design provided? ‐ The research question and inclusion criteria should be established before the conduct of the review | Yes | Yes |
| 2. Was there duplicate study selection and data extraction? | No | Yes |
| 3. Was a comprehensive literature search performed? | Incomplete ‐ CENTRAL and MEDLINE were searched with 4 different search terms (no search strings) on 20 April 2010. This electronic database search was not supplemented by screening reference lists and contacting researchers/clinicians in the field | Yes |
| 4. Was the status of publication (i.e. grey literature) used as an inclusion criterion? | Can't answer, nothing about publication and language was reported | No |
| 5. Was a list of studies (included and excluded) provided? ‐ A list of included and excluded studies should be provided | No ‐ A list of excluded studies was not reported | Yes |
| 6. Were the characteristics of the included studies provided? | No ‐ A list of included studies was provided but it did not contain all relevant information (e.g. characteristics of participants, interventions, control) | Yes |
| 7. Was the scientific quality of the included studies assessed and documented? | No | Yes |
| 8. Was the scientific quality of the included studies used appropriately in formulating conclusion? | No | Yes |
| 9. Were the methods used to combine the | No ‐ 2 meta‐analyses were performed, 1 that was accompanied by a I2 value of 89.8% and 1 did not report an I2 value. The term 'heterogeneity' did not appear in the article. In addition, the software used to conduct the meta‐analyses was not reported and nothing mentioned about the effects model (i.e. fixed or random) used | Yes |
| 10. Was the likelihood of publication bias assessed? | No ‐ Publication bias was not mentioned in the article | Not applicable ‐ We set out in our protocol to assess the likelihood of publication bias, but since we had too few studies per comparison we could not draw a funnel plot |
| 11. Was the conflict of interest stated? | Yes ‐ The author declared that she had no competing interests and she stated that there was no funding | Yes |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Recovery Show forest plot | 3 | 599 | Risk Ratio (M‐H, Random, 95% CI) | 1.32 [1.16, 1.50] |
| 1.1 HIV‐uninfected and untested children | 2 | 534 | Risk Ratio (M‐H, Random, 95% CI) | 1.32 [1.10, 1.58] |
| 1.2 HIV‐infected children | 1 | 65 | Risk Ratio (M‐H, Random, 95% CI) | 1.41 [0.97, 2.04] |
| 2 Relapse Show forest plot | 3 | Risk Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2.1 HIV‐uninfected and untested children | 2 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 2.2 HIV‐infected children | 1 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.0 [0.0, 0.0] | |
| 3 Mortality Show forest plot | 3 | 599 | Risk Ratio (M‐H, Random, 95% CI) | 0.97 [0.46, 2.05] |
| 3.1 HIV‐uninfected and untested children | 2 | 534 | Risk Ratio (M‐H, Random, 95% CI) | 0.78 [0.32, 1.88] |
| 3.2 HIV‐infected children | 1 | 65 | Risk Ratio (M‐H, Random, 95% CI) | 1.69 [0.42, 6.85] |
| 4 Weight gain (g/kg/day) Show forest plot | 3 | 595 | Mean Difference (IV, Random, 95% CI) | 1.47 [0.49, 2.45] |
| 4.1 HIV‐uninfected and untested children | 2 | 530 | Mean Difference (IV, Random, 95% CI) | 1.79 [0.65, 2.93] |
| 4.2 HIV‐infected children | 1 | 65 | Mean Difference (IV, Random, 95% CI) | 0.80 [‐0.64, 2.24] |
| 5 Time to recovery for HIV‐uninfected children (days) Show forest plot | 1 | 136 | Mean Difference (IV, Fixed, 95% CI) | ‐7.0 [‐15.89, 1.89] |
| 6 Mid‐upper arm circumference gain (mm/day) Show forest plot | 3 | 570 | Mean Difference (IV, Random, 95% CI) | 0.13 [0.04, 0.21] |
| 6.1 HIV‐uninfected and untested children | 2 | 505 | Mean Difference (IV, Random, 95% CI) | 0.15 [0.07, 0.24] |
| 6.2 HIV‐infected children | 1 | 65 | Mean Difference (IV, Random, 95% CI) | ‐0.04 [‐0.28, 0.20] |
| 7 Weight for height z score at follow‐up in HIV‐uninfected children Show forest plot | 1 | 99 | Mean Difference (IV, Fixed, 95% CI) | 0.19 [‐0.22, 0.60] |
| 8 Days of diarrhoea during the intervention period Show forest plot | 1 | 352 | Mean Difference (IV, Fixed, 95% CI) | ‐0.60 [‐1.30, 0.10] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Recovery Show forest plot | 2 | 210 | Risk Ratio (M‐H, Random, 95% CI) | 0.71 [0.60, 0.84] |
| 1.1 HIV‐uninfected children | 1 | 162 | Risk Ratio (M‐H, Random, 95% CI) | 0.72 [0.60, 0.87] |
| 1.2 HIV‐infected children | 1 | 48 | Risk Ratio (M‐H, Random, 95% CI) | 0.62 [0.39, 0.99] |
| 2 Relapse Show forest plot | 2 | 210 | Risk Ratio (M‐H, Random, 95% CI) | 8.95 [1.18, 67.77] |
| 2.1 HIV‐uninfected children | 1 | 162 | Risk Ratio (M‐H, Random, 95% CI) | 15.25 [0.91, 255.90] |
| 2.2 HIV‐infected children | 1 | 48 | Risk Ratio (M‐H, Random, 95% CI) | 5.07 [0.28, 93.00] |
| 3 Mortality Show forest plot | 2 | 210 | Risk Ratio (M‐H, Random, 95% CI) | 0.73 [0.25, 2.18] |
| 3.1 HIV‐uninfected children | 1 | 162 | Risk Ratio (M‐H, Random, 95% CI) | 0.48 [0.08, 2.81] |
| 3.2 HIV‐infected children | 1 | 48 | Risk Ratio (M‐H, Random, 95% CI) | 0.95 [0.24, 3.80] |
| 4 Weight gain (g/kg/day) Show forest plot | 2 | Mean Difference (IV, Fixed, 95% CI) | Subtotals only | |
| 4.1 HIV‐uninfected children | 1 | 158 | Mean Difference (IV, Fixed, 95% CI) | ‐2.10 [‐3.08, ‐1.12] |
| 4.2 HIV‐infected children | 1 | 48 | Mean Difference (IV, Fixed, 95% CI) | ‐0.10 [‐1.73, 1.53] |
| 5 Time to recovery for HIV‐uninfected children (days) Show forest plot | 1 | 116 | Mean Difference (IV, Fixed, 95% CI) | 10.0 [0.87, 19.13] |
| 6 Mid‐upper arm circumference gain (mm/day) Show forest plot | 2 | 173 | Mean Difference (IV, Random, 95% CI) | ‐0.11 [‐0.22, 0.01] |
| 6.1 HIV‐uninfected children | 1 | 125 | Mean Difference (IV, Random, 95% CI) | ‐0.15 [‐0.27, ‐0.03] |
| 6.2 HIV‐infected children | 1 | 48 | Mean Difference (IV, Random, 95% CI) | ‐0.03 [‐0.20, 0.14] |
| 7 Weight for height z score at follow‐up for HIV‐uninfected children Show forest plot | 1 | 73 | Mean Difference (IV, Fixed, 95% CI) | ‐0.10 [‐0.56, 0.36] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Recovery Show forest plot | 1 | 1874 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.97 [0.93, 1.01] |
| 2 Relapse Show forest plot | 1 | 1874 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.33 [1.03, 1.72] |
| 3 Mortality Show forest plot | 1 | 1874 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.90 [0.55, 1.45] |
| 4 Weight gain (g/kg/day) Show forest plot | 1 | 1874 | Mean Difference (IV, Fixed, 95% CI) | ‐0.5 [‐0.75, ‐0.25] |
| 5 Mid‐upper arm circumference gain (mm/day) Show forest plot | 1 | 1874 | Mean Difference (IV, Fixed, 95% CI) | ‐0.04 [‐0.06, ‐0.02] |
| 6 Weight for height z score Show forest plot | 1 | 1874 | Mean Difference (IV, Fixed, 95% CI) | 0.0 [‐0.10, 0.10] |
| 7 Weight for age z score Show forest plot | 1 | 1874 | Mean Difference (IV, Fixed, 95% CI) | ‐0.10 [‐0.21, 0.01] |
| 8 Height for age z score Show forest plot | 1 | 1874 | Mean Difference (IV, Fixed, 95% CI) | ‐0.10 [‐0.24, 0.04] |
